Production and 3D Printing Processing of Bio-Based Thermoplastic Filament

Manufacturing Rev. 2017, 4,1 E. Gkartzou et al., Published by EDP Sciences, 2017 DOI: 10.1051/mfreview/2016020 Available online at: http://mfr.edp-open.org

RESEARCH ARTICLE OPEN ACCESS

Production and 3D printing processing of bio-based thermoplastic ﬁlament

Eleni Gkartzou, Elias P. Koumoulos, and Costas A. Charitidis* Research Unit of Advanced, Composite, Nano Materials & Nanotechnology, National Technical University of Athens, School of Chemical Engineering, 9 Heroon Polytechniou St., Zographos, Athens 15780, Greece

Received 22 July 2016 / Accepted 22 October 2016

Abstract – In this work, an extrusion-based 3D printing technique was employed for processing of biobased blends of Poly(Lactic Acid) (PLA) with low-cost kraft lignin. In Fused Filament Fabrication (FFF) 3D printing process, objects are built in a layer-by-layer fashion by melting, extruding and selectively depositing thermoplastic fibers on a platform. These fibers are used as building blocks for more complex structures with defined microarchitecture, in an automated, cost-effective process, with minimum material waste. A sustainable material consisting of lignin biopolymer blended with poly(lactic acid) was examined for its physical properties and for its melt processability during the FFF process. Samples with different PLA/lignin weight ratios were prepared and their mechanical (tensile testing), thermal (Differential Scanning Calorimetry analysis) and morphological (optical and scanning electron microscopy, SEM) properties were studied. The composition with optimum properties was selected for the production of 3D-printing filament. Three process parameters, which contribute to shear rate and stress imposed on the melt, were examined: extrusion temperature, printing speed and fiber’s width varied and their effect on extrudates’ morphology was evaluated. The mechanical properties of 3D printed specimens were assessed with tensile testing and SEM fractography.

Key words: 3D Printing processability assessment, Fused ﬁlament fabrication, Additive manufacturing, Biobased 3D printing ﬁlament

1. Introduction dealing with the main drawbacks of lignin usage regarding the low-purity standards, heterogeneity, smell and color prob- In the last decade, issues concerning environmental pollu- lems of the existing commercial lignins. It is recognized that tion and the increasing awareness of limited resources, have blending lignin with polymers is a convenient and inexpensive motivated the scientific community to study and optimize method to create new materials with tailored properties, such renewable alternatives to traditional petroleum-derived plastics, as hydrophobicity, stiffness, crystallinity, thermal stability, like biobased composite materials that are sourced from car- Ultraviolet (UV) blocking ability and to reduce the overall cost bon-neutral feedstocks [1]. Lignin is a highly aromatic of the material [4, 5]. biopolymer, abundantly found in the fibrous part of various The ecosystem of 3D printing plastic market consists of plants and extracted as a byproduct of wood pulping industries. numerous research and development activities and is projected Kraft lignin (sulfate lignin) is isolated in the so-called deligni- to reach USD 692.2 Million by 2020, at a Compound Annual fication process, from black liquor by precipitation and neutral- Growth Rate (CAGR) of 25.7% from 2015 to 2020 [6]. Among ization with an acid solution (pH = 1–2), and subsequently the various commercially available specialty filaments for driedtoasolidform[2]. It is estimated that only 2% of the Fused Filament Fabrication (FFF) processing, materials mim- industrially extracted lignin is exploited for low-volume, niche icking wood texture and properties are a separate category, applications, while the rest is often relegated to a low efficiency because of their ability to create objects with the tactile feel energy recovery via combustion or as a natural component of of wood without any need for specialized woodworking tools. animal feeds [3]. Thus, the development of ways to convert lig- Furthermore, they require less maintenance and preservatives, nin to new high-value products is an active area of research, since they are more resistant to organic decomposition, while maintaining their biodegradability. PLA/Lignin (poly(lactic *e-mail: [email protected] acid)/Lignin) 3D printing filaments are an alternative option

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 2 E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1 for lignin exploitation and filament cost reduction and can be form the successive cross sections of the physical part. In most used in rapid prototyping, presentation models and consumer CAM programs for lower end FFF 3D printers, VFR is a func- products, among others. Poly(lactic acid) is a biodegradable tion of the linear feed velocity of the filament and of several thermoplastic, which is produced via fermentation or chemical design parameters related to the toolpath (e.g. the width and synthesis from a bio-derived monomer, lactic acid (2-hydroxy height of individual fibers, defined by extrusion width and propionic acid) [7]. The carbon in PLA originates from atmo- layer height parameters) [14]. The pressure-driven mass flow spheric carbon dioxide, which is immobilized in glucose by of the non-Newtonian polymeric melt through the nozzle is photosynthesis; therefore, its impact on the environment during mainly related to nozzle geometry, pressure drop and melt’s production and disposal (carbon footprint) is low compared to apparent viscosity. This flow can be described as a fully devel- other petro-based polymers. PLA is widely used in 3D printing oped, laminar flow through a capillary die with a circular cross applications, since it is one of the most user-friendly materials section [14, 15]. The necessary pressure for fiber extrusion is that can be easily processed with FFF, without emitting toxic applied by the solid portion of the 3D printing filament which fumes. However, its low thermal stability, high degradation rate acts as a piston, as it is pushed by a pinch-roller feed mecha- during processing and brittle behavior have to be addressed. It nism into a melting reservoir, placed on the upper part of the has been suggested that the presence of lignin increases ther- nozzle. Volumetric flow rate along with extrusion temperature mal stability and flammability under oxidative and nonoxida- are two material-dependent factors which contribute on the tive conditions, due to the formation of char, which acts as a shear rate and stresses imposed on the melt during extrusion. protective layer preventing oxygen diffusion [8]. In the case of composite 3D printing materials, these factors Extrusion-based 3D printing techniques use temperature as are significantly influenced and limited by the filler’s dispersion a way of controlling the material state, for the successful extru- and agglomeration [16]. sion and deposition of semi-molten thermoplastic fibers, on a This study is divided in two main sections; the first section flat surface. In a typical FFF process, a filament feedstock is involves the preparation and characterization of bulk samples supplied to the system by an electric motor-controlled pinch of the composite material with increasing lignin content and roller mechanism [9, 10]. Material is liquefied inside a reser- 3D printing filament production. The second section concerns voir, contained in a heated metal block with a machined chan- the selection of suitable toolpath and process parameters based nel, so that it can flow through the print head’s nozzle and fuse on the produced filament’s response during FFF processing, with adjacent material before solidifying. This approach is focusing on extrusion temperature, print head’s velocity and similar to conventional polymer extrusion processes, except extrusion width. By understanding the relationship between the extruder is vertically mounted on a plotting system (print processing conditions and physical phenomena involved in head) rather than remaining in a fixed horizontal position the material’s extrusion and deposition, suitable bounds for [10]. After its deposition, the solidifying material is referred these parameters were derived. The optimum velocity and tem- to as a fiber or road. The part is produced by superimposing perature values were used for the fabrication of tensile speci- a specified number of layers, where each of them is generated mens with 100% nominal density and three different by a specific pattern of fibers. The formation of bonds among extrusion widths, produced by three brass nozzles with differ- individual fibers in the FFF process consists of complicated ent diameter (0.2, 0.3 and 0.4 mm), in order to measure the heat and mass transfer phenomena coupled with thermal and tensile properties of the finished parts and to compare them mechanical stress accumulation and phase changes. The with the properties of the bulk material. Pure PLA filament strength of these bonds depends on the growth of the neck was produced and processed under the same conditions, to formed between adjacent fibers and on the molecular diffusion be used for comparison. Fractographic analysis of tensile fail- and randomization at the interface [11]. As a natural conse- ure was carried out with Scanning Electron Microscopy quence of this manufacturing approach, the part’s internal (SEM). Also, a qualitative assessment of the filler’s dispersion microstructure consists of fibers with partial bonding among and agglomeration into the polymeric matrix, as well as its them and voids [12] and can be assimilated to a composite effect on surface morphology and diameter of individual fibers two-phase material with inherently orthotropic properties was made with optical microscopy. [13]. Individual fibers are significantly stronger in the axial direction and resemble the fibers in a composite; however, the structure shows weaker behavior in the direction where 2. Experimental details stresses need to be carried through fiber-to-fiber or layer-to- layer adhesion [12]. 2.1. Materials Three dimensional printers employing the FFF technology are Computer Numerical Control (CNC) machines, whose Commercial PLA pellets under the grade name ‘‘INGEO function is defined by a program containing coded alphanu- 2003D’’ were supplied by Natureworks LLC, with number meric data (G-code). G-code sets of commands are typically average molecular weight Mn (g/mol) = 114.317, weight aver- generated by Computer-Aided Manufacturing (CAM) soft- age molecular weight Mw (g/mol) = 181.744 and 4.3 wt.% ware, which uses topological information from 3D Com- D-isomer content [17]. A purified form of kraft pine lignin puter-Aided Design (CAD) data along with user-defined (Indulin AT) was supplied by MWV Specialty Chemicals, in processing and toolpath parameters, to create virtual slices of the form of free-flowing powder with a wide distribution of the object to be manufactured and to calculate the toolpath particle diameter, as depicted in the SEM micrographs of and the material’s Volumetric Flow Rate (VFR), in order to Figure 1. Both materials were used without chemical treatment E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1 3

(a) (b)

Figure 1. SEM micrographs of kraft lignin (Indulin AT) particles (a) ·800 and (b) ·1.600 magnification. for the preparation of the blends. Prior to processing, the com- Table 1. Sample composition. ponents were vacuum-dried at 50 C for 24 h and weighted in a high precision scale. Sample PLA wt.% Lignin wt.% LPLA00 100 0 LPLA05 95 5 2.2. Blending LPLA10 90 10 LPLA15 85 15 Firstly, 40 g samples of PLA/Lignin blends with different LPLA20 80 20 lignin concentrations (5, 10, 15, 20% percentage by weight on the dry polymer – Table 1) were prepared by melt mixing in a twin-screw Brabender internal mixer. Mixing time for each were performed at room temperature with a Zwick tensile tes- sample varied from 10 to 13 min, at 35 rpm rotating speed of ter, model 1120 equipped with a 2000 N load cell. Both the screws and mixing temperature between 180–190 C (de- Young’s modulus and elongation measurements were made at pending on lignin content). To remove residual stresses and a constant crosshead speed of 2 mm/min. Each value of air bubbles caused by the blending process, the semi-molten mechanical properties reported is an average of five specimens. material was collected from the mixing chamber and placed DSC analyses were performed with DSC Q200 TA Instruments in an aluminum mold, which was inserted in a Dake thermo- (New Castle, DE, USA). The thermal history of samples was press to form a 15 · 60 · 1 mm plate. The plates of the ther- erased by a preliminary heating cycle, followed by a cooling mopress were heated at 120 C and heating was switched off cycle from 200 to 0 C and a second heating cycle from 0 when maximum load was applied on the mold. All samples to 200 C. Both cooling and heating rates were set at 10 C/ used for the bulk material’s characterization were cut from min. The samples’ mass ranged from 6.90 to 8.36 mg and they the aforementioned plates and kept under room temperature were encapsulated in aluminum pans. An empty pan was in a glass desiccator, to prevent moisture absorption. used as reference. The glass transition temperature (Tg) cold crystallization temperature (Tcc), double melting peak temper- 2.3. Characterization atures (Tm1,2), cold crystallization enthalpy (DHcc) and melting enthalpy (DHm) were determined from heating scans. Thermo- The effect of increasing lignin content on the bulk mate- gravimetric Analysis (TGA) of softwood kraft lignin was car- rial’s thermal and mechanical properties was evaluated with ried out based on global mass loss with a Netzsch 409 EP Differential Scanning Calorimetry (DSC) and tensile testing. analyzer, from which lignin’s decomposition pattern can be The phase morphology of the samples was examined with an derived. The analysis was conducted under nitrogen atmo- Axio Imager A2m optical microscope and AxioCam ICc5 sphere with a heating rate of 10 C/min. The characterization CCD camera (Carl Zeiss, Oberkochen, Germany) and SEM of 3D printed fibers and tensile specimens involved optical micrographs were taken with Nova NanoSEM 230 scanning microscopy and tensile testing. Brightfield illumination was electron microscope (FEI Company) with an acceleration volt- used to observe the surface roughness of PLA/Lignin fibers. age of 5 kV. Samples for optical inspection were cut with a Also, by exploiting PLA’s transparency and alterations in the rotating saw and embedded in cold mounting epoxy resin. incident light’s state of polarization during its interaction with The embedded samples’ surface was grinded with fine silicon lignin, a qualitative evaluation of the dispersion of lignin’s carbide abrasives to remove defects introduced by sectioning. agglomerates in the fibers’ bulk volume was conducted. Axio- Tensile specimens were directly cut from the thermopressed Visio digital image processing software was used to measure plates of the compounded material with a dumbbell-shaped the diameter of the fibers from the micrographs captured by specimen cutting die, with a 18 · 3 · 1 mm reduced gage sec- the CCD camera. Since no special standard exists for the char- tion. Measurements of mechanical properties of specimens acterization of FFF parts, tensile specimens based on a scaled 4 E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1 down version of ASTM D638 Type I with a 36 · 8 · 2mm gauge section, were 3D printed using Zmorph 2.0 S, a commercial FFF Cartesian XZ system (Zmorph LLC, Wroclaw, Poland). Tensile testing was carried out under the same conditions as the bulk material specimens, with the same crosshead speed of 2 mm/min. Both pure PLA and composite PLA/ Lignin filaments were used as raw materials. The 3D printed specimens had 100% infill density in order to make a comparison with the tensile properties of the bulk material and to eliminate errors from G-code generation. Divergence from the nominal dimensions of the tensile specimens introduced by the fabrication process, was measured for each specimen with a high precision digital caliper and the average of 10 measured values for each dimension was used for stress calcula- tions. Fractographic analysis of tensile failure was carried out with scanning electron microscopy. The fractured samples were sputter-coated with a thin layer of gold before observation. Figure 2. Diameter distribution of filament sections used for specimen fabrication.

2.4. Production of 3D printing filament The 3D printing filament needs to be able to provide and sustain the pressure needed to drive the extrusion process. Failure to do this results in filament buckling, which occurs when the extrusion pressure is higher than the critical buckling load that the filament can support [10, 14, 18]. The filament’s elastic modulus determines its load carrying ability and melt viscosity determines the resistance to extrusion (or extrusion pressure). As a result, the composition with 5 wt.% lignin content was selected for the production of 3D printing filament with 1.75 mm nominal diameter, which is compatible with the 3D printer’s feeding system. Subsequently, a Boston-Mathews single-screw extruder with an L:D ratio of 25:1 with a 1.8 mm diameter extrusion die was used to obtain PLA/Lignin filament and several combinations of process parameters were tested in order to achieve enhanced filler dispersion and constant filament diameter. A processing temperature profile of 185–195–205–205–195 C from feed section to die was selected along with a fixed screw speed at 16 rpm. The extru- date was collected by a conveyor belt equipped with cooling fans, whose speed was matched to the extrusion speed in order to control the diameter of the filament. The composite filament and pure PLA filament (processed under a 180–190–200–200–190 C temperature profile) were used to obtain 3D printed specimens and individual fibers. The pro- Figure 3. Illustrations of toolpath for rectilinear infill pattern with 90 raster angle generation and cross sections of the produced duced filament was conditioned in room temperature inside pattern and individual roads, where W is the extrusion width, H is sealed plastic bags, in the presence of silica gel, to avoid mois- r r layer height and d is raster-to-raster distance. ture absorption. Filament sections with tight tolerances were r selected for specimen fabrication. Each section’s diameter was measured every 1 cm with a high precision digital caliper and average diameter and standard deviation were calculated. head’s liquefier can be considered as a two-dimensional, From Figure 2 it can be seen that the measured filament axisymmetric, steady-state, advection-conduction heat transfer diameters are normally distributed around the mean value, process [19]. Gaps between the filament and the wall of the with 0.02 mm standard deviation, meaning that 95% of the liquefier, caused by the filament’s diameter inconsistency can filament used has a ±0.04 tolerance. A tight diameter be expected to hinder heat transfer and result in irregular tolerance is important during the material melting and viscous behavior of the melt on the upper part of the liquefier. deposition process. The heating of the filament inside the print Diameter inconsistencies have adverse effects on the material E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1 5

Table 2. Fused filament fabrication toolpath and process parameters. Toolpath parameters Process parameters Layer height (mm) 0.1 Speed for non-print moves (mm/s) 100 Filament diameter (mm) 1.78 ± 0.04 Vertical shells (#) 3 Speed for print moves (mm/s) 20 Extrusion multiplier 1 Horizontal shells (#) 0 Infill/perimeters overlap (mm) 0.15 Extrusion temperature (C) 205 Seam position Aligned First layer speed (mm/s) 10 Building platform heating Off Infill density (%) 100 Extrusion width (mm) Equal to nozzle diameter Fan speed (%) 20 Infill pattern Rectilinear Infill/perimeters overlap (mm) 0.1 Disable fan for the first 2 Layers Infill angle (degree) 90 Minimum detail resolution (mm) 0.02

feeding mechanism as well, since mismatches between the new layers, new physical contacts are generated; hence, several roller and filament surfaces may lead to filament slipping. heat transfer modes change and heat transfer with air entrapped Furthermore, the filament’s mean diameter is used by the Com- between contiguous filaments may also develop. Toolpath and puter Aided Manufacturing (CAM) program, which automati- G-code generation have a significant effect on thermal stress cally calculates the material’s feed velocity in the 3D printer’s accumulation in fibers and layers and thus different CAM pro- extruder. As a result, the diameter’s standard deviation is grams with the same input values, produce parts with different related to volumetric flow rate fluctuations during the 3D print- responses to external stresses. In this study, an open-source ing process, which can alter the distance between adjacent G-code generating program (Slic3r 1.2) was used for specimen fibers, causing insufficient bonding or fiber overlapping and fabrication. A second G-code generating program (Voxelizer thus reducing the physical object’s dimensional accuracy and 1.4), with a different generating algorithm, was used with the structural integrity. same input values of toolpath and process parameters, in order to estimate the effect of different CAM programs on the mechanical properties of the final part. 2.5. Computer aided manufacturing – toolpath As far as process parameters are concerned, they and process parameters include extrusion temperature, chamber temperature, cooling rate, filament feed velocity and volumetric flow rate, among Computer aided manufacturing software typically enables others. The extrusion process does not have a considerable control of several design and process parameters related to influence on the strength and modulus of the material, but fused filament fabrication. Design parameters define the tool- notably affects the maximum strain, since during extrusion path followed by the nozzle’s tip and include individual fibers’ through the nozzle, polymer chains are submitted to stress- width and height (commonly referred to as ‘‘road/extrusion induced orientation, which reduces the elongation characteris- width’’ and ‘‘layer’s height/thickness’’, respectively), as well tics of the material [12]. The rate at which the filament is fed to as various deposition strategies to form and fill the successive the liquefier (feed velocity) is dynamically controlled and con- cross sections of the part. A common deposition strategy is to nected to velocity changes of the print head, in order to main- separately deposit one or more continuous contours of all tain a constant volumetric flow rate. The amount of melt which boundary 2D surfaces included in a given cross section and is present in the reservoir, the temperature of the melt, and con- to fill the space between them, (corresponding to the part’s sequently, the viscosity and surface energy of the melt, vary interior), with specific infill patterns. Part orientation in relation with feed rate [14]. In the case of constant linear movement to infill orientation and to the system’s main axes of movement of the print head, the extruder motor speed is proportionate (XYZ for cartesian 3D printers) plays an important role in sur- to printing speed, so an indirect control of feed speed can be face finish, dimensional accuracy, cost and mechanical behav- achieved varying printing speed for the same toolpath charac- ior [20, 21]. In the simple case of rectilinear infill pattern teristics. Extrusion temperature and deposition rate are identi- (Figure 3), each layer is filled with a raster of parallel roads fied as the major parameters influencing inter- and intra-layer and adjacent layers have a fixed 90 raster angle between them. bonding [24]. By adjusting infill density, which is expressed as a percentage User-defined processing parameters for G-code generation of occupied space, more sparse or dense parts can be produced and specimen fabrication are listed in Table 2, which resulted with bigger or smaller distances among contiguous fibers of from temperature and printing speed optimization, as well as the same layer (raster to raster distance, dr). A slightly negative adjustments on the standard processing profile for PLA recom- dr, corresponding to fiber overlapping, has been reported to mended by the manufacturer. A constant layer thickness of reduce void density and increase contact area among fibers 0.1 mm was used for all specimens and extrusion widths were and thus resulting in stronger fiber-to-fiber bonds [22, 23]. determined by nozzle diameter. Three nozzles with the same However, the excessive material buildup at the layer’s perime- design characteristics and 0.2, 0.3, 0.4 mm openings were ter significantly affects dimensional accuracy on the XY plane. tested and extrusion width was set equal to the respective noz- As individual fibers are deposited on the previously solidified zle diameter. Nozzle diameter, along with the viscosity of the layer of the material, heat exchanges by conduction develop melt, determine the pressure drop during extrusion and thus the on contact surfaces between adjacent fibers and by convection force required from the feed mechanism. Pressure drop and radiation with the surroundings. Upon the deposition of increases as nozzle diameter increases, or as feed velocity 6 E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1

Figure 4. Toolpath simulation for tensile specimen fabrication with three contours and rectilinear inﬁll pattern with 100% nominal density. Specimens consist of alternating Type A and Type B layers.

Figure 5. (a–b) Reflected light (brightfield and polarized illumination) micrographs of PLA/Lignin with 5% lignin (magnification ·100 and ·500). (c–d) Closer examination (·4000 and ·8000 magnification) of the morphology of lignin aggregation with SEM, from the tensile- fractured surface of sample with 15% lignin. increases. Furthermore, in the vicinity of the nozzle, the poly- same position near the center of the build platform. Toolpath mer melt is under stress and part of the deformation energy simulation generated by G-code generating software is stored elastically leads to radial expansion of the fiber after depicted in Figure 4. Before the fabrication of each specimen, extrusion. Individual fibers were extruded from each nozzle a touch probe sensor (by Zmorph LLC) was used for precise at three print speeds (20, 40 and 60 mm/s). Surface morphol- leveling of the build platform and the nozzle’s distance from ogy and diameter were studied at the center of the fibers, in surface (0.1 mm) was verified at four points within the surface order to avoid errors related to print head acceleration and of specimen fabrication. deceleration. Since no standard test specimens exists for characterization of parts processed with FFF, tensile specimens were fabricated based on a 3D model of ASTM D638 Standard 3. Results Test Method for Tensile Properties of Plastics, Type I speci- 3.1. Bulk material characterization men, designed with Autodesk Fusion 360 CAD and extracted in fine quality Stereolithography (STL) format. The original 3.1.1. Reflected light microscopy specimen was scaled down by ·0.6 to avoid fabrication errors near the boarders of the available working volume. To ensure The morphological analysis of binary PLA/Lignin blends the repeatability of the fabrication process, all tensile speci- was performed with reflected light microscopy and scanning mens were fabricated separately, with their wide surface paral- electron microscopy. In general, the morphology is defined lel to the XY plane and gauge section parallel to X axis, at the by the complex thermomechanical history experienced by the E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1 7

Figure 6. Engineering stress-strain curves of the PLA/Lignin blends with 0, 5, 10 and 15 wt.% lignin content. Figure 7. Percentage change of PLA’s mechanical properties (ultimate tensile strength, Young’s modulus of elasticity and Table 3. Tensile properties of PLA/Lignin composites. elongation at break) with increasing lignin content. Sample E (GPa) UTS (MPa) el (%) LPLA00 2.31 ± 0.04 55.9 ± 0.6 4.57 ± 0.22 lignin has no adverse effect on Young’s modulus of elasticity, LPLA05 2.33 ± 0.05 50.3 ± 0.9 2.81 ± 0.10 asitcanbeseeninFigure 6 and Table 3. Samples with 5 LPLA10 2.41 ± 0.06 50.1 ± 0.5 2.32 ± 0.17 and 10 wt.% lignin displayed a 40% and 49% reduction in LPLA15 2.39 ± 0.06 41.3 ± 0.5 1.88 ± 0.34 elongation at break respectively. The relative change in PLA’s Young’s modulus of Elasticity (E), Ultimate Tensile Strength (UTS) and Elongation at Break (el) with increasing lignin con- different constituents during processing [25]. All samples tent is presented in Figure 7. For samples LPLA10 and formed heterogeneous systems, due to the low compatibility LPLA15, UTS is equal to breaking strength (absence of yield between PLA matrix and the unmodified kraft lignin, which point), while the breaking strength of pure PLA and PLA with has been previously reported [24–28]. At 5 wt.% lignin con- 5 wt.% lignin is slightly smaller than UTS, at 48.1 and tent, the morphology mainly consists of a uniform dispersion 47.6 MPa, respectively. The sample with 20 wt.% lignin con- of lignin aggregates of small size (<20 lm) in homoge- tent was too brittle for tensile specimens to be cut. neous surrounding matrix of PLA/Lignin, as it can be seen at Figures 5a and 5b. At higher lignin content, there is an 3.1.3. Thermal characterization increase in aggregates’ concentration, which remained uni- formly dispersed. At 20% lignin, particle size increases The DSC thermograms recorded during cooling and sec- because of coalescence phenomena. ond heating of PLA composites are reported in Figure 8.Asin- A closer examination of lignin aggregation at smaller gle glass transition temperature (Tg) registered for all blends, scales can be achieved from SEM analysis of the fractured sur- but no deduction can be made for the miscibility of the ingre- face of the samples (Figures 5c and 5d). Prior to examination, dients, since lignin’s content is low and its Tg is possibly over- sample surfaces were covered with gold sputter coating. It can lapped by PLA’s melting peaks. At the chosen cooling rate be seen that lignin aggregates consist of smaller lignin domains (10 C/min), the material did not develop a significant crys- of fairly circular shape, closely packed in specific regions of talline component and no crystallization peaks were recorded the PLA matrix. Interfacial separation and sliding between lig- during cooling cycle both for pure PLA and PLA/Lignin com- nin’s aggregates and the surrounding matrix are indicative of posites. Therefore, a highly amorphous structure is obtained very weak secondary interactions between the two polymers. with thermodynamic instability, which can form local ordered structure during annealing below glass transition temperature 3.1.2. Tensile testing due to the enthalpic relaxation of the polymeric chains [29, 30], the melting of which is manifested as a small endothermic The type of morphology and the phase dimensions deter- peak during the heating cycle after the glass transition process mine the mechanical and physical properties of the blend at (Figure 8). When annealed between glass transition large, which is influenced by intermolecular lignin-lignin, temperature (Tg) and melting temperature (Tm), the mobility PLA-PLA and PLA-lignin interactions. Consequently, the of chain segments of glassy polymers increases gradually observed immiscibility between the two constituents results and cold crystallization occurs [31]. Lignin appears to promote in minor effective stress transfer between lignin’s aggregates PLA’s double melting behavior, which is attributed to the melt- and PLA matrix and increases PLA’s brittleness, which is ing of two populations of lamellae. The lower temperature shown as a significant reduction in the plastic region of peak is connected to the melting of small lamellae produced stress-strain curves and disappearance of yield point. However, by secondary crystallization, while the higher temperature 8 E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1

(a) (b)

Figure 8. (a) DSC thermographs of PLA/lignin bulk composites. (b) DSC thermograph of sample with 5 wt.% lignin. Quantities for characterization of the glass transition of sample containing 5 wt.% lignin: extrapolated onset temperature (Tge), half-step temperature (Tg1/2), change of the normalized heat capacity during transition (DCp), initial (Tgi) and ﬁnal (Tgf) temperatures of the glass transition.

Table 4. Transition temperatures and associated changes in heat capacity, enthalpy and crystallinity of PLA/Lignin composites.

À1 À1 Sample Tgi (C) Tgf (C) Tg1/2 (C) Tge (C) DCp (JC g ) Tcc (C) Tm1 (C) Tm2 (C) DHcc (J/g) DHm (J/g) LPLA00 47.4 71.0 59.2 56.0 0.52 113.4 149.2 153.9 À25.0 25.9 LPLA05 49.0 69.0 57.4 55.7 0.48 113.6 148.9 154.3 À24.7 25.3 LPLA10 48.0 70.0 59.0 55.0 0.46 116.3 149.3 154.2 À24.4 25.0 LPLA20 45.5 69.0 57.3 54.7 0.43 111.8 148.2 155.0 À22.3 22.4 peak originates from the melting of major crystals formed in the primary crystallization process. As part of the ongoing research related to crystalline behavior of semi-crystalline polymers, it has been suggested that the primary crystals are first formed, followed by the formation of secondary crystals and simultaneous thickening of the primary crystals [32]. Other authors have explained double melting behavior with a melt-recrystallization model, where the low-temperature and high-temperature peaks in the DSC curve are attributed to the melting of some amount of the original crystals and to the melting of crystals formed through a melt-recrystallization process during the heating scan, respectively [33]. As lignin content increases, there is a greater contribution from the melting of major crystals to the specific melting enthalpy of the sample or from crystals formed through a melt-recrystallization process, which is implied by the increase in the higher temperature melting peak in Figure 8. Therefore, for the given Figure 9. TGA plot of Indulin AT kraft lignin and the derivative heating rate lignin affects nucleation and crystal growth, possi- TGA curve, obtained under nitrogen atmosphere at 10 C/min bly encouraging recrystallization or the formation of thicker heating rate. crystalline structures during the cold crystallization process. Basic quantities which characterize the samples’ glass transition, cold crystallization and melting process were derived for cold crystallization (Tcc), low melting peak (Tm1), high from DSC thermographs and are listed in Table 4.Nosignifi- melting peak (Tm2) and the related cold crystallization and cant changes were observed for the initial (Tgi), final (Tgf), melting enthalpies also seem to be independent from lignin extrapolated onset (Tge) and half-step (Tg1/2) glass transition content, with very small variations. As a result, no special temperature and for the associated change in heat capacity intermolecular interaction between PLA and lignin can be con- (DCp) before and after glass transition. The peak temperature firmed, in accordance to the observed phase morphology. E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1 9

Figure 10. Macrographs of temperature calibration specimen captured with Sony SLT-A57 SAL-1855, depicting changes on surface roughness caused by thermal degradation (in black/white contrasted images, from darker to brighter images.

Table 5. Characteristic decomposition temperatures of Indulin AT kraft lignin.

Sample To Tend T1 T2 T3,max T5% T30% T50% Kraft lignin 216 511 101.6 156.5 372.8 237 379 606

Furthermore, since the FFF process involves melt quenching at every 1 cm across Z axis (10 layers with 0.1 mm height). higher cooling rates and instant solidification of the material A brass nozzle with 0.4 mm diameter was used for specimen after its deposition, it can be deducted that the material’s crys- fabrication and the rest of the user-defined input values for tallinity during and after FFF processing will mainly be defined the CAM program are listed in Table 2. It has been observed by stress-induced crystallization. that the 3D printing extrusion temperature is generally higher The Thermogravimetry (TG) (in weight loss percentage) than that needed for filament extrusion [34], due to the mate- and Derivative Thermogravimetry (DTG) (in weight loss per- rial’s short residence time in the liquefier’s reservoir and the centage per C) curves of Indulin AT kraft lignin, obtained limited power of the feeding system’s stepper motor (NEMA at 10 C/min heating rate under nitrogen atmosphere, are plot- 17), compared to the single-screw extruder. Also, the liquefier ted in Figure 9. Thermal degradation data indicates weight loss temperature set on CAM program reflects the temperature of and the first derivative (DTG) indicates the corresponding rate the melt reservoir on the upper part of the nozzle, close to of weight loss. The peak of DTG can be presented as a mea- the print head’s temperature sensor. The correct deposition sure of thermal decomposition and can be used as a means temperature is expected to be lower, since the tip of the brass to compare thermal stability characteristics of different materi- nozzle itself is not heated and insulated [35]. From the images als. As it can be seen in Figure 10, thermal decomposition of Figure 10, it can be seen that surface roughness is directly occurred over a wide temperature range, starting at approxi- connected to the material’s thermal stability. At temperatures mately 216 C(To). Lignin’s weight loss percentages for vari- higher than 215 C, the anisotropy of surface’s topography ous characteristic decomposition temperatures are summarized increases, in accordance to lignin’s thermal decomposition pat- in Table 5,whereTo is initial decomposition temperature, Tend tern (Figure 9). At 190 C, the melt’s viscosity is significantly is terminal decomposition temperature, T1 and T2 are attributed higher, and lignin’s agglomerates hinder material flow, result- to the evolution of humidity and chemically bound water ing in nozzle clogging. The optimum combination of surface respectively, T3,max is maximum decomposition temperature roughness and material flow was achieved at 205 C. corresponding to DTG peak and T5%, T30%, T50% are temperatures of 5, 10, 30 and 50% weight loss due to degradation. 3.2.2. Extrusion of individual fibers and infill inspection

3.2. Fused filament fabrication process Both PLA and PLA with 5 wt.% lignin fibers with 15 cm length were individually extruded from the FFF system at optimization 205 C, using nozzles with different openings (0.4, 0.3, 3.2.1. Extrusion temperature 0.2 mm). From Figures 11 and 12, it is evident that lignin causes a severe increase in fibers’ surface roughness (in com- A temperature calibration specimen, consisting of a parison with pure PLA fibers) and the actual extrusion width 20 · 20 · 80 mm hollow rectangular cuboid with 0.8 cell is mainly defined by the size and distribution of lignin aggre- thickness, was modeled in Autodesk Fusion 360 CAD program gates. Phase separation is more intense in filament as opposed and extracted in STL file format, in order to assess the impact to the bulk material with the same lignin content, due to the of liquefier temperature on surface finish (Figure 10). Extru- dispersive mixing limitations of the single-screw system sion temperature varied from 230 to 190 C, with 5 Csteps used for filament production, as to the twin-screw internal 10 E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1

Figure 11. Individual ﬁbers extruded from 0.2 mm nozzle (a–b) 20 mm/s printing speed and (c–d) 60 mm/s, where lignin aggregation causes up to 30% increase in ﬁber’s diameter, respectively.

Figure 12. Reflected light (brightfield and polarized illumination) micrographs of specimen with 50% rectilinear infill density and three successive contours, for inspection of lignin dispersion, surface texture and volumetric flow fluctuations. compounder used for sample preparation. For the tested inconsistencies become more prominent at higher speeds, as print velocities (20, 40 and 60 mm/s), no nozzle clogging it can be seen in Figures 11c and 11d, where the local diameter occurred. However, as nozzle diameter decreases and becomes increases by 30%. Therefore, 20 mm/s speed was selected as a comparable with the size of lignin’s aggregates, diameter common printing speed for all nozzles. A cylindrical specimen E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1 11

(a) (b)

(c)

Figure 13. (a) Effect of different nozzle diameter on specimen’s tensile properties. Comparison with the tensile properties of bulk material with the same (5 wt.%) lignin content. (b) Effect of different CAM programs on tensile properties of specimens with the same toolpath and process parameters. Representative examples with and without premature internal failure. (c) Tensile properties of specimens produced from pure PLA and PLA/Lignin ﬁlament for two different nozzle diameters.

with 1 cm radius and 1 cm height, with 3 contours and its standard deviation from average value), because of the rectilinear infill pattern (50% nominal density) was fabricated imponderable effect of the increased stress-induced orientation with 0.4 mm extrusion width for infill inspection. Uneven of polymer chains [12, 21]. The 0.4 mm diameter nozzle distribution of lignin phases of various sizes is depicted at resulted in specimens with maximum elongation at break sim- Figure 12, along with a case of over-extrusion dew to ilar to that exhibited by bulk material. Ultimate tensile strength volumetric flow fluctuations during the extrusion of the contour and Young’s modulus of elasticity have similar values for dif- of the specimen. ferent road widths and are 18% and 6% reduced respectively, compared to bulk material. Figure 13b and Table 7 present 3.2.3. Tensile properties of 3D printed specimens the tensile properties of representative specimens fabricated with two different G-code generation programs and the same As it has been noted by other studies [12, 13], small vari- user-defined process and toolpath parameters. Two characteris- ations of the processing conditions can strongly affect speci- tic responses to tensile stresses appeared in both sets of tensile mens mechanical properties. The reported values and specimens, indicating that specimens fabricated under the standard deviations are the average of five specimens, which same conditions can exhibit different fracture behaviors, failed within the specimen’s gauge section. The effect of differ- because of premature inter- and intra-laminar failures related ent extrusion widths is presented in Figure 13a and Table 6. to over- or under-extrusion or weak bonds among individual As it has been mentioned, extrusion through different nozzles fibers, especially at the fusion points between the infill and negatively affects specimen’s maximum strain (and increases the internal contour. This can be also observed with SEM 12 E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1

Table 6. Tensile properties PLA/Lignin 3D printed specimens with various extrusion widths – comparison with the bulk material. Sample Extrusion width (mm) E (GPa) UTS (MPa) el (%) PLA/Lignin filament 0.2 2.18 ± 0.06 41.8 ± 1.6 2.31 ± 0.16 PLA/Lignin filament 0.3 2.20 ± 0.18 40.2 ± 0.7 2.55 ± 0.19 PLA/Lignin filament 0.4 2.16 ± 0.10 43.6 ± 2.1 2.83 ± 0.26 PLA/Lignin bulk material – 2.33 ± 0.05 50.3 ± 0.9 2.81 ± 0.10

Table 7. Tensile properties of representative PLA with 5 wt.% lignin specimens, with and without premature failure, exhibited for both CAM programs used for fabrication.

Sample CAM program E (GPa) Spf (MPa) Elpf (%) UTS (MPa) el (%) Vox_PLA/Lignin a Voxelizer 2.03 22.4 1.30–1.34 39.2 2.99 Vox_PLA/Lignin b Voxelizer 2.16 – – 43.6 2.93 Sli_PLA/Lignin a Slic3r 2.25 24.0 1.23–1.28 43.1 2.88 Sli_PLA/lignin b Slic3r 2.17 – – 41.8 2.44

Figure 14. SEM micrographs of the fractured surface of 3D printed tensile specimens. The upper/lower corners of gauge region cross section consist of three successive contours. PLA/Lignin a and PLA/Lignin b, which were fabricated with different CAM programs, exhibit different fracture morphologies. analysis of the fractured surfaces of tensile specimens fabri- approximately half of the corresponding ultimate tensile cated with different CAM programs (Figure 14 – 1a–1c and strength and is noted as Spf in Table 7.Elpf is the strain region 2a–2c). The stress around which premature failure occurs is corresponding to negative slope of stress-strain curve. E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1 13

Table 8. Tensile properties of specimens fabricated from pure PLA and PLA/Lignin ﬁlament. Sample composition Extrusion width (mm) E (GPa) UTS (MPa) el (%) PLA 0.3 2.17 ± 0.12 48.2 ± 0.9 3.68 ± 0.35 PLA with 5 wt.% Lignin 0.3 2.21 ± 0.33 40.9 ± 0.9 2.58 ± 0.16 PLA 0.4 2.13 ± 0.08 43.6 ± 1.7 2.93 ± 0.21 PLA with 5 wt.% Lignin 0.4 2.19 ± 0.10 41.9 ± 1.1 2.44 ± 0.18

Figure 13c and Table 8 present a comparison of tensile processing will mainly be defined by stress-induced crystalliza- properties among identical specimens fabricated from pure tion. Lignin appears to promote PLA’s double melting behavior PLA and PLA/lignin filaments. As the tensile stress increases, and affects nucleation and crystal growth, possibly encourag- the failure will begin at the weakest raster and next weakest ing recrystallization or the formation of thicker crystalline raster will break, in sequence, until total failure of the sample structures during the cold crystallization process. Thermogravi- [36]. As expected, lignin addition mainly reduces elongation at metric analysis of kraft lignin indicates that its thermal decom- break and increases PLA’s brittle behavior, which can be also position occurred over a wide temperature range, starting at deducted by comparing the characteristics of PLA’s fractured approximately 216 C. Since increasing lignin content results surface (Figure 14 – 3a–3c) with PLA/Lignin specimens. in severe agglomeration and increase in the material’s brittle- PLA’s interlayer fusion at the infill region appears to be ness and melt’s apparent viscosity during extrusion, the compo- improved every two layers, which is caused by toolpath varia- sition with 5 wt.% lignin content was selected for the tion (Figure 4), since the fabrication of Type-A layers has a production of 3D printing filament and segments of reduced time between successive depositions of fibers and 1.78 ± 0.04 mm diameter were used for extrusion of individ- temperature at the bonding site remains higher for a longer ual fibers and fabrication of tensile specimens. The addition period of time, improving neck formation and growth with of lignin causes a severe increase in fibers’ surface roughness the preceding layer [37, 38]. Interestingly, this effect is limited (in comparison with pure PLA fibers) and the actual extrusion in PLA/Lignin specimens, where all layers seem to have width is mainly defined by the size and uneven distribution of similar bond quality, with more complex fracture mechanisms. lignin aggregates. Based on the material’s response during pro- The non-selective trans-layer and trans-fiber crack propagation cessing, the optimum extrusion temperature (205 C) and is indicative of improved bond quality, which could be printing speed (20 mm/s) for extrusion through nozzles with explained by the adhesion-enhancing effect of lignin addition, different diameters (0.2, 0.3 and 0.4 mm) were used for tensile caused by the increase in surface roughness of individual fibers specimen fabrication. Extrusion through different nozzles neg- [38]. As it has been mentioned, crack initiation and propaga- atively affects specimen’s maximum strain (and increases its tion is more probable to occur in specimen’s upper/lower standard deviation from average value), because of the impon- corners and more specifically in contour region, where void derable effect of the increased stress-induced orientation of density is higher, or at the infill-contour fusion points. For polymer chains. The 0.4 mm diameter nozzle resulted in spec- the chosen process and toolpath parameters, very good imens with maximum elongation at break similar to that exhib- adhesion among contiguous fibers and successive layers was ited by bulk material. Ultimate tensile strength and Young’s achieved. modulus of elasticity have similar values for different road widths and are 18% and 6% reduced respectively, compared to bulk material. It was also found that specimens fabricated 4. Conclusions under the same conditions can exhibit different fracture behaviors, because of premature inter- and intra-laminar failures In this study, the properties of biobased PLA/Lignin blends related to over- or under-extrusion or weak bonds among indi- were examined and correlated to processing conditions with vidual fibers, especially at the fusion points between the infill fused filament fabrication and tensile properties of the pro- and the internal contour. The non-selective trans-layer and duced parts. Samples with different PLA/lignin weight ratios trans-fiber crack propagation observed in SEM micrographs were prepared and their phase morphology as well as mechan- of the fractured surface is indicative of improved bond quality, ical and thermal properties were examined. It was found that which could be explained by the adhesion-enhancing effect all samples formed heterogeneous systems, where interfacial caused by the increase in surface roughness of individual separation and sliding between lignin’s aggregates and the sur- PLA/Lignin fibers compared to pure PLA. Additionally, spec- rounding PLA matrix are indicative of very weak secondary imens fabricated with different CAM programs with the same interactions between the two polymers. Phase separation also input values for basic toolpath and processing parameters, results in minor effective stress transfer between lignin’s aggre- exhibit different fracture morphologies, which implies that gates and PLA matrix, which increases PLA’s brittleness and G-code generation algorithms also affect the final part’s causes a significant reduction in the plastic deformation region mechanical properties. of stress-strain curves and disappearance of yield point. Future studies will focus on employing Design Of Experi- However, lignin has no adverse effect on Young’s modulus of ment (DOE) techniques for systematic analysis and determina- elasticity. Furthermore, the material’s slow crystallization rate tion of the relation among basic toolpath and processing (at 10 C/min cooling rate no crystallization peaks were parameters and output measures, aiming for an overall mate- recorded), suggests that its crystallinity during and after FFF rial-based optimization of the FFF process. As far as the raw 14 E. Gkartzou et al.: Manufacturing Rev. 2017, 4,1 material is concerned, various methods for improving PLA 12. A. Bellini, S. Güçeri, Rapid Prototyping Journal 9 (2003) compatibility with kraft lignin will be tested (addition of plas- 252–264. ticizers and compatibilizing agents, use of lignin fractions with 13. M. Bertoldi, M.A. Yardimici, C.M. Pistor, S.I. Güçeri, G. Sala, lower polydispersity indices, lignin chemical treatment) and Mechanical characterization of parts processed via fused their effect on the material’s dynamic rheological properties deposition, in: Solid Freeform Fabrication Proceedings, Austin, will be examined. Also, thermal stability under oxidative and TX, 1998. nonoxidative conditions, as well as thermal degradation during 14. B.N. Turner, R. Strong, S.A. Gold, Rapid Prototyping Journal processing will be investigated. 20 (2014) 192–204. 15. N. Sa’ude, M. Ibrahim, M.H.I. Ibrahim, Applied Mechanics Acknowledgements. This work has partially received funding from and Materials 660 (2014) 89–93. the European Union’s Horizon 2020 research and innovation pro- 16. N. Mostafa, et al., Tsinghua Science and Technology 14 (2009) gram MODCOMP under grant agreement No 685844 and EU 29–37. FP7 Project ‘‘Functionalized Innovative Carbon Fibres Developed 17. J.F.M. de Almeida, et al., Polymer Bulletin 73 (2016) from Novel Precursors with Cost Efficiency and Tailored Proper- 3531–3545. ties’’ (FIBRALSPEC) under Grant Agreement No. 604248. 18. N. Venkataraman, et al., Rapid Prototyping Journal 6 (2000) 244–253. References 19. L.B. Ji, T.R. Zhou, Advanced Materials Research 97–101 (2010) 2585–2588. 1. F.G. Calvo-Flores, et al., Applications of modified and 20. M. Domingo-Espin, et al., Materials & Design 83 (2015) unmodified lignins, in: Lignin and lignans as renewable raw 670–677. materials, John Wiley & Sons Ltd, New York City, United 21. B.N. Turner, S.A. Gold, Rapid Prototyping Journal 21 (2015) States, 2015, pp. 247–288. 250–261. 2. H. Chung, N.R. Washburn, 2 – Extraction and types of lignin, 22. A.K. Sood, R.K. Ohdar, S.S. Mahapatra, Journal of Advanced in: Lignin in polymer composites, William Andrew Publishing, Research 3 (2012) 81–90. New York City, United States, 2016, pp. 13–25. 23. M. Dawoud, I. Taha, S.J. Ebeid, Journal of Manufacturing 3. J.A.G. Ochoa de Alda, Resources, Conservation and Recycling Processes 21 (2016) 39–45. 52 (2008) 965–972. 24. S.F. Costa, F.M. Duarte, J.A. Covas, Virtual and Physical 4. H. Naegele, et al., 13 – Applications of lignin materials and Prototyping 10 (2014) 35–46. their composites (lignin applications in various industrial 25. R. Dell’Erba, et al., Polymer 42 (2001) 7831–7840. sectors, future trends of lignin and their composites), in: 26. C. Pouteau, et al., Comptes Rendus Biologies 327 (2004) Lignin in polymer composites, William Andrew Publishing, 935–943. New York City, United States, 2016, pp. 233–244. 27. J. Li, Y. He, Y. Inoue, Polymer International 52 (2003) 5. Y.-L. Chung, et al., ACS Sustainable Chemistry & Engineering 949–955. 1 (2013) 1231–1238. 28. O. Gordobil, et al., Polymer Degradation and Stability 108 6. MarketsandMarkets, 3D printing plastic market by type (pho- (2014) 330–338. topolymers, abs, pla, polyamide/nylon, others), by form (fila- 29. R. Surana, et al., Thermochimica Acta 433 (2005) 173–182. ment, ink, powder), by application, by end-user industry, and by region – global forecasts to 2020, 2016, Available from: http:// 30. J.M. Hutchinson, Progress in Polymer Science 20 (1995) www.marketsandmarkets.com/Market-Reports/3d-printing- 703–760. plastic-market-21707470.html. 31. C. Zhou, et al., Polymer 90 (2016) 111–121. 7. H. Tsuji, Poly (lactic acid), in: Bio-based plastics, John Wiley 32. L. Baldenegro-Perez, et al., Polymers 6 (2014) 583–600. & Sons Ltd, New York City, United States, 2013, pp. 171–239. 33. M. Yasuniwa, et al., Journal of Polymer Science Part B: 8. O. Faruk, et al., 6 - Lignin Reinforcement in Thermoplastic Polymer Physics 42 (2004) 25–32. Composites, in Lignin in Polymer Composites, William 34. A. Melocchi, et al., International Journal of Pharmaceutics 509 Andrew Publishing, New York City, United States, 2016, pp. (2016) 255–263. 95–118. 35. S.I.G, S.C.D., M. Atif Yardimci, in: Solid Freeform Fabrication 9. A. Bellini, S.U. Güçeri, M. Bertoldi, Journal of Manufacturing Symposium Proceedings, Austin, TX, 1997. Science and Engineering 126 (2004) 237. 36. W. Wu, et al., Materials 8 (2015) 5834–5846. 10. I. Gibson, W.D. Rosen, B. Stucker, Extrusion-Based Systems, 37. Y.K. Chou, Y. Zhang, Proceedings of the Institution of in: Additive manufacturing technologies: rapid prototyping to Mechanical Engineers Part B: Journal of Engineering direct digital manufacturing, Boston, MA, Springer US, 2010, Manufacture 220 (2006) 1663–1671. pp. 160–186. 38. F. Awaja, et al., Progress in Polymer Science 34 (2009) 11. C. Bellehumeur, et al., Journal of Manufacturing Processes 6 948–968. (2004) 170–178.

Cite this article as: Gkartzou E, Koumoulos EP & Charitidis CA: Production and 3D printing processing of bio-based thermoplastic filament. Manufacturing Rev. 2017, 4,1.